Item – Thèses Canada

​
Numéro d'OCLC
66890422
Lien(s) vers le texte intégral
Exemplaire de BAC
Exemplaire de BAC
Auteur
Pelletier, Denis,1974-
Titre
Problems in time series and financial econometrics : linear methods for VARMA modelling, multivariate volatility analysis, causality and value-at-risk.
Diplôme
Ph. D. -- Université de Montréal, 2004
Éditeur
Ottawa : Library and Archives Canada = Bibliothèque et Archives Canada, [2005]
Description
3 microfiches.
Notes
Includes bibliographical references.
Résumé
The objective of this thesis is to study various problems in time series and financial econometrics. The common thread of the various parts is the intrinsic curse of dimensionality underlying the study of multivariate time series. In the first essay, we consider the problem of modelling VARMA models by relatively simple methods which require linear regressions. For that purpose, we consider the regression-based estimation method proposed by Hannan and Rissanen (1982, 'Biometrika') for univariate ARMA models. The asymptotic properties of the estimator are derived under weak hypotheses for the innovations (uncorrelated and strong mixing) so as to broaden the class of models to which it can be applied. To further ease the use of VARMA models we present new identified VARMA representations, 'diagonal MA equation form' and 'final MA equation form', where the MA operators are diagonal and scalar respectively. We also present a modified information criterion which gives consistent estimates of the orders of these representations. To demonstrate the importance of using VARMA models to study multivariate time series we compare the impulse-response functions generated by VARMA and VAR models. In the second essay, we propose a new model for the variance between multiple time series, the Regime Switching Dynamic Correlation model. In this model, we decompose the covariances into correlations and standard deviations. The correlation matrix follows a regime switching model: it is constant within a regime but different across regimes. The transitions between the regimes are governed by a Markov chain. This model does not suffer from a curse of dimensionality and it allows analytic computation of multi-step ahead conditional expectations of the variance matrix. We also present an empirical application which illustrates that our model can have a better in-sample fit of the data than the Dynamic Conditional Correlation model proposed by Engle (2002, ' JBES'). In the third essay, we discuss methods for testing hypothesis of non-causality at various horizons, as defined in Dufour and Renault (1998, 'Econometrica '). We study in detail the case of VAR models and we propose linear methods based on running vector autoregressions at different horizons. While the hypotheses considered are nonlinear, the proposed methods only require linear regression techniques as well as standard Gaussian asymptotic distributional theory. For the case of integrated processes, we propose extended regression methods that avoid nonstandard asymptotics. The methods are applied to a VAR model of the U.S. economy. In the fourth essay, we propose new statistical tests for backtesting financial risk models used for computing Value-at-Risk (VaR), like the model we proposed in the second essay. These tests are based on the duration in days between the violations of the VaR. Our Monte Carlo results show that in realistic situations, the new duration-based tests have considerably better power properties than the previously suggested tests.
ISBN
0612979075
9780612979079